1/ Managers particularly for SDE roles don’t usually have >10-12 people. In reality the URA (Unregretted Attrition) goal is mostly managed at the L8+ level where organization sizes are usually hundreds. Far more likely you have some underperformers in an overall organization of 300, even if an individual team is all awesome.
2/ Interviews at Amazon involve a Bar Raiser who almost certainly won’t work in the Hiring Managers span of control, and the debrief attempts to answer “Is this a bar raising hire?”, meaning is this applicant better than 50% of people in the role/level.
I can really only speak for AWS and Engineering roles, but I never saw any “Hire to Fire” behavior at Amazon, and most every Bar Raiser that I saw took the role seriously and was not a “rubber stamper”.
Balls may have been dropped by some people, but I always have recruiters pinging me relentlessly the day after I've done an interview so they can get back to the candidate within 5 business days.
You are a Director; you have a 350 person organization; you’re leading a rapidly growing business (let’s say >40% YoY growth in revenue, not uncommon in parts of AWS); during the OP1 you secured incremental investment of 20% meaning you’ll grow to 420, which is +70 hires to make.
You’re goaled against a 6% URA meaning 21 people exit voluntarily after being entered to Focus, during Pivot, or involuntarily at the unsuccessful end of a Pivot.
If I spoke from experiences, I’d say the (vast) majority of your URA will likely come from the tenure group with < 18 months at Amazon, in effect, hiring decisions which didn’t work out and where, in debrief, everyone thought the person had potential to be a bar raising hire but in reality they were bottom quartile. In the debrief your clear options were “Hire at L7”, “Hire at L6”, “No Hire”, plus maybe some flavor of good fit for Amazon but not really for this specific role.
Let’s take the extreme of this and say all URA comes from just the +70 growth, 21 people would be a 30% failure rate.
In a less extreme and more likely scenario: it’s more like some fast failures occur from the +70, others are slightly slower and show up after 12-18 months (that would be the prior years growth). Tenured people with e.g. >4 years sometimes start to struggle for whatever reason.
That’s not all that unexpected, in my opinion! Even with a structured well-used interviewing mechanism you’re getting perhaps 6 hours of signal and during this time the applicants are naturally very focused on “putting the best foot forward”. After a hire decision you work with someone for 6 months or more and ask “Is this what I expected?”, and “If I could make the decision again based on the signal I now possess, would I still be strongly convicted and proceed to hire this person?”. When the answers to these questions are “No”, that’s often what commences the performance management processes.
Now, this explanation works for growth businesses, it’s less apparent how 6% URA makes sense in a flat growth organization, since you’d be finding those cuts from a mostly tenured and presumably mostly proven set of people.
